comment on

I'm benchmarking the latest SQLite for use in sharing data between mod_perl processes. The use case is a shared hash, which is modeled as a single table with columns "a_key" and "value". However, my code is running pretty slow. Simple MySQL code doing the same thing goes three times as fast, and BerkeleyDB goes about 17 times as fast. I'm wondering if SQLite is really this slow, or if I'm just not aware of some important tuning step with it. Removing the commit after storing data makes it three times as fast, but then it's no longer doing something comparable to the MySQL and BDB code since other processes would not be able to see that data.

The SQLite part of my test code is below. I just bang on this with a couple of hundred thousand hits to store() and fetch() and see how long it takes. I'm using the latest DBI and DBD::SQLite.

package IPC::SharedHash::DBDSQLite;

use strict;
use warnings;

use DBI;
use DBD::SQLite;
use Storable qw(nfreeze thaw);

my $DB_NAME = 'SQLite';

sub new {
    my $class   = shift;
    my $self    = {};
    my %options = @_;
    $self->{'SCALARS_ONLY'} = 1;

    my $dbh =
      DBI->connect_cached(
        'dbi:SQLite:dbname=' . $options{'DIRECTORY'} . '/' . $DB_NAME,
        '', '', { AutoCommit => 0 })
      || die;
    $dbh->{'RaiseError'} = 1;
    $dbh->do('PRAGMA synchronous = OFF');
    unless ( $dbh->tables( '', '', 'shared_hash', '' ) ) {
        $dbh->do(
            'create table shared_hash ( a_key text primary key, value 
+text )');
    }

    my $fetch_sth =
      $dbh->prepare('select value from shared_hash where a_key = ?');

    my $store_sth =
      $dbh->prepare('replace into shared_hash (a_key, value) values (?
+, ?)');

    $self->{'dbh'}   = $dbh;
    $self->{'fetch'} = $fetch_sth;
    $self->{'store'} = $store_sth;
    
    bless $self, $class;
    return $self;
}

sub fetch {
    my $self = shift;
    my ($key) = @_;

    $self->{'fetch'}->execute($key);
    my $value;
    $self->{'fetch'}->bind_columns( \$value );
    $self->{'fetch'}->fetch();
    $self->{'fetch'}->finish();
    if ( $self->{'SCALARS_ONLY'} ) {
        return $value;
    } else {
        if ( defined $value ) {
            return thaw($value);
        }
    }
}

sub store {
    my $self = shift;
    my ( $key, $value ) = @_;
    if ( !$self->{'SCALARS_ONLY'} ) {
        warn $self->{'SCALARS_ONLY'};
        $value = nfreeze($value);
    }

    $self->{'store'}->execute( $key, $value );
    $self->{'dbh'}->commit();
}

sub DESTROY {
   my $self = shift;
   $self->{'dbh'}->disconnect();
}

1;
[download]

... and here's a small sample of calling it:

#!/usr/bin/perl

use strict;
use warnings;

use IPC::SharedHash::DBDSQLite;

my $hash = IPC::SharedHash::DBDSQLite->new(
    DIRECTORY    => '/tmp',
    SCALARS_ONLY => 1,
);

for ( 0 .. 10000 ) {
    $hash->store( "key$_", $_ );
    my $return = $hash->fetch("key$_");
    die "bad return: $return" unless $return == $_;
}
[download]

In reply to DBD::SQLite tuning by perrin

Are you posting in the right place? Check out Where do I post X? to know for sure.
Posts may use any of the Perl Monks Approved HTML tags. Currently these include the following:
<code> <a> <b> <big> <blockquote> <br /> <dd> <dl> <dt> <em> <font> <h1> <h2> <h3> <h4> <h5> <h6> <hr /> <i> <li> <nbsp> <ol> <p> <small> <strike> <strong> <sub> <sup> <table> <td> <th> <tr> <tt> <u> <ul>
Snippets of code should be wrapped in <code> tags not <pre> tags. In fact, <pre> tags should generally be avoided. If they must be used, extreme care should be taken to ensure that their contents do not have long lines (<70 chars), in order to prevent horizontal scrolling (and possible janitor intervention).
Want more info? How to link or How to display code and escape characters are good places to start.


There's more than one way to do things
	PerlMonks